Week 10: Local RAG

Objective

Build a simple retrieval-augmented system locally so you can see how searching notes before answering can improve quality.

Required Videos

• One short RAG explainer from a trusted source
• Optional: compare the explainer to what the script actually does

Tasks

1. Run python week10_rag.py.
2. Ask at least three questions about the notes.
3. Inspect which note chunks were retrieved.
4. Add one more note section and test again.

Deliverables

• At least three successful question-and-answer runs
• One example showing retrieved context
• One short explanation of why retrieval helps

Checkpoint Questions

• What is retrieval doing before generation?
• Why is this simpler than a production RAG system?
• What happens when the right keywords are missing?
• How would you improve the retriever?